refactor(parser): check go pkg name using regex

Cargo.toml

@@ -14,6 +14,8 @@ categories = ["development-tools"]
 [dependencies]
 clap = { version = "4.5.4", features = ["derive"] }
 ignore = "0.4.22"
+once_cell = "1.19.0"
+regex = "1.10.4"
 schemars = "0.8.16"
 serde = { version = "1.0.197", features = ["derive"] }
 serde_json = "1.0.115"

src/lib.rs

@@ -4,13 +4,15 @@ use config::MdsfConfig;
 use error::MdsfError;
 use formatters::format_snippet;
 use languages::Language;
+use parser::{parse_generic_codeblock, parse_go_codeblock};
 use terminal::{print_changed_line, print_file_info, print_line_info, print_unchanged_file};
 
 pub mod cli;
 pub mod config;
 pub mod error;
 pub mod formatters;
 pub mod languages;
+mod parser;
 pub mod runners;
 pub mod terminal;
 
@@ -33,32 +35,15 @@ fn format_file(config: &MdsfConfig, input: &str) -> (bool, String) {
         // TODO: implement support for code blocks with 4 `
         if line.starts_with("```") {
             if let Some(language) = line.strip_prefix("```").and_then(Language::maybe_from_str) {
-                let mut code_snippet = String::new();
-
-                let mut is_snippet = false;
-
-                let mut snippet_lines = 0;
-
-                for (_, subline) in lines.by_ref() {
-                    snippet_lines += 1;
-
-                    if subline.trim_end() == "```" {
-                        is_snippet = true;
-                        break;
-                    }
-
-                    code_snippet.push_str(subline);
-
-                    code_snippet.push('\n');
-                }
+                let (is_snippet, code_snippet, snippet_lines) = if language == Language::Go {
+                    parse_go_codeblock(&mut lines)
+                } else {
+                    parse_generic_codeblock(&mut lines)
+                };
 
                 if is_snippet {
                     print_line_info(language, line_index + 1, line_index + snippet_lines + 1);
 
-                    if language == Language::Go && !code_snippet.contains("package ") {
-                        code_snippet.insert_str(0, GO_TEMPORARY_PACKAGE_NAME);
-                    }
-
                     let formatted = format_snippet(config, &language, &code_snippet);
 
                     output.push_str(line);

src/parser/mod.rs

@@ -0,0 +1,79 @@
+use core::{iter::Enumerate, str::Lines};
+
+use once_cell::sync::Lazy;
+use regex::Regex;
+
+use crate::GO_TEMPORARY_PACKAGE_NAME;
+
+#[inline]
+pub fn parse_generic_codeblock(lines: &mut Enumerate<Lines>) -> (bool, String, usize) {
+    let mut code_snippet = String::new();
+
+    let mut is_snippet = false;
+
+    let mut snippet_lines = 0;
+
+    for (_, subline) in lines.by_ref() {
+        snippet_lines += 1;
+
+        if subline.trim_end() == "```" {
+            is_snippet = true;
+            break;
+        }
+
+        code_snippet.push_str(subline);
+
+        code_snippet.push('\n');
+    }
+
+    (is_snippet, code_snippet, snippet_lines)
+}
+
+#[inline]
+pub fn parse_go_codeblock(lines: &mut Enumerate<Lines>) -> (bool, String, usize) {
+    let (is_snippet, mut code_snippet, snippet_lines) = parse_generic_codeblock(lines);
+
+    if is_snippet && !GO_PACKAGE_RE.is_match(&code_snippet) {
+        code_snippet.insert_str(0, GO_TEMPORARY_PACKAGE_NAME);
+    }
+
+    (is_snippet, code_snippet, snippet_lines)
+}
+
+// TODO: check for multiline comments
+pub static GO_PACKAGE_RE: Lazy<Regex> = Lazy::new(|| Regex::new(r"^\s*package\s+\w").unwrap());
+
+#[cfg(test)]
+mod test_go_package_re {
+    use crate::parser::GO_PACKAGE_RE;
+
+    #[test]
+    fn it_should_match() {
+        for s in [
+            "package\tmdsf",
+            "package mdsf ",
+            "  package mdsf",
+            "\n package mdsf",
+            "\n package mdsf\t",
+            "\n package    mdsf",
+            "\n package \tmdsf",
+            "\n package\tmdsf",
+            "\n \tpackage\t\n\nmdsf\n",
+        ] {
+            assert!(GO_PACKAGE_RE.is_match(s), "'{s}' did not match");
+        }
+    }
+
+    #[test]
+    fn it_should_not_match() {
+        for s in [
+            "packageasd",
+            "missing pkg name",
+            "// package mdsf ",
+            "//package mdsf",
+            "//\tpackage mdsf",
+        ] {
+            assert!(!GO_PACKAGE_RE.is_match(s), "'{s}' matched");
+        }
+    }
+}